When attempting to reboot or shutdown a system that has no local disks or
minimal local disks (e.g. swap and /var local, everything else NFS-mounted), the
shutdown process stops network filesystems and network interfaces before all
other shutdown steps are complete.
There should be a way, perhaps if a check of /etc/mtab shows / or other critical
filesystems to be mounted via NFS , to defer shutdown of NFS and related
services, as well as the network interface in question until the very end.
Steps to Reproduce:
1. Configure a system so that its root filesystem is mounted via NFS instead of
2. After booting this system, attempt to reboot it or shut it down using the
typical 'reboot' or 'shutdown' commands.
Error messages that result from a premature unmounting of the very filesystem
that contains rc .
E.g. following the "Bringing <if> down..." sequence
/sbin/rc: line <N>: <command>: Network unreachable
Execute the whole shutdown process cleanly, leaving NFS and the network
interface(s) to be shut down at the end.
neglected to post `emerge info` like bug report page said to
(In reply to comment #1)
> neglected to post `emerge info` like bug report page said to
You're right, my mistake. This was on an AMD64 system setup via 2005.0 stage3
LiveCD, with no system updates applied and any additional required packages
After I performed `emerge baselayout` , the rc scripts were updated to latest
version, and this type of situation *is* correctly handled by current stable
baselayout. Latest stable startup scripts checkroot and checkfs also handle NFS
root situation much more gracefully than 2005.0 stage3 (no /fastboot file needed).
This bug / flaw has already been addressed. Marking bug "CLOSED". Apologies
for the waste of bandwidth.